In this post, Virginia Prevosto,FCAS and Peter Marotta, AIDM from ISO provide an account of how best-practice techniques ensures high levels of data quality across billions of insurance premium records.

Profit by Data Quality Best Practices 

For Want of a Nail the Kingdom was Lost

Like the missing horseshoe nail that cost a kingdom in the old Mother Goose nursery rhyme, faulty data can produce devastating bottom line consequences in the property / casualty insurance business.

Consider how incomplete or miscoded data can produce cascading miscalculations in underwriting, risk selection, coverage and pricing. Data is at the heart of customer service and support functions. Claims data — from both internal and external sources — support decisions and planning for claims settlement. Past claims experience help companies identify emerging claim trends. The quality of data can lead to either profitably helpful or dangerously misleading predictions for future claims-adjustment and settlement costs.

Insurers use data to manage litigation, detect fraudulent claims and limit financial exposure to claims through reinsurance, but this practice works only when the data is credible. It is no overstatement that sound, profitable property / casualty operations begin – and end – with quality data.

But what is quality data and how can companies attain and sustain that quality?

Our firm, whose expertise in data aggregation and management is widely recognized by the insurance industry, has developed best practices and standards to ensure the quality of our statistical data. We maintain one of the largest private databases in the world, with 10.6 billion records at any given time. Although we have developed these principles for the insurance business, the elements and principles that follow are equally applicable to all organizations.

We define quality data as data fit for its intended use. The five key characteristics of quality data are:

• Accuracy: information in the database represents exactly what it is supposed to
  capture.
• Validity: the value of a data element in the database is identified as an allowable
  value.
• Reasonability: data is consistent with prior data or other available information.
• Completeness: every recorded transaction contains all necessary information, and all
  pertinent transactions are being recorded and reported.
• Timeliness: transactions are consistently recorded, processed and changed within
  established and prescribed timeframes.

With the above characteristics of data quality in mind, here are the key principles we recommend as guidelines to organizations for managing data quality:

Data Stewardship

• Maintain a corporate program with senior management oversight
• Understand roles and responsibilities in data ownership, acquisition, quality assurance, storage and distribution
• Make each functional area with data responsibility accountable for their own performance and data management

Data and Data Quality Standards

• Develop internal standards and, where appropriate, seek useful external standards
• Harmonize multiple standards and promote consistent operations across multiple systems and platforms

Organizational Issues 

• Establish an in-house unit to create and assess data
• Aquisition across the organization
• Tap into external resources where appropriate

Operations and Processes

• Develop processes to maximize data quality and use new technologies to manage data
• Monitor regulatory requirements that may affect data and data quality

Data Element Development and Specification

• Design and maintain data, system, and reporting mechanisms for sound data management and quality for end-user service
• Review the current level of data detail and assess whether or not historical or retrospective data are necessary for developing system or reporting specifications
• Define data element and design data-reporting specifications to enable convenient modifications and updates

Data Management and Data Quality Tools

• Develop tools such as a corporate data dictionary, edits and business rules, data-flow documentation, process model and mapping, and data-translation criteria by data source and recipient
• Leverage technology resources like the Internet, predictive and data-visualization tools, and new data exchange standards for improved data management and quality

Measurement

• Develop performance metrics to measure poor data quality, such as costs for crrecting errors and reports, investigating and preventing errors, fines and regulatory scrutiny.
• Benchmark results for each data source

Individual support

• Institute support for data management and data quality at both individual and organizational levels

Privacy

• Educate users about privacy issues, policies and compliance with privacy regulations
• Control access to, and use of, non-public data
• Promote best practices in data privacy

In Summary

Just as organizations protect their financial, physical and intellectual property assets, so should organizations ensure that their data assets are accurate, reliable and protected from unauthorized access.

As businesses in various industries "go global", the value of data quality for informed business decisions, financial success, and market reputation will also grow exponentially.

Does your organization adopt any of the techniques in this article? What are your experiences? Please share your views below. 

Did you find this article useful?

If so, please tweet it: "Profit by Data Quality Best Practices" http://bit.ly/2Hkd1w #dataquality #datagovernance

Useful Resources

See all posts in: Methodology

• Does Your Business Suffer From a Data Quality Reality Gap?
• How to set data quality goals any business can achieve
• How To Create A Data Issue Assessment Process: Expert Interview With Ken O'Connor
• How to create a data quality competency center: Expert interview with John Schmidt
• Creating An Internal Data Quality Community: Introduction (Part 1 of 4)
• Information Quality Management Framework (IQMF): An Overview
• Embedding Data Quality in Business Intelligence Reports: Introductory Tutorial
• 15 Tips for transforming knowledge-workers into a data quality task force
• How to create a data quality framework or data quality methodology:Essential resources to get you started

About the Authors

Virginia Prevosto and Peter Marotta are principals in the Consulting and Research Department of ISO (http://www.iso.com), based in Jersey City, N.J., maintains one of the largest private databases in the world by obtaining roughly 2 billion detailed records of insurance premiums collected and losses paid each year.

ISO’s professional staff members analyze insurer data and turn it into meaningful information. ISO provides data, analytics and decision-support solutions to professionals in many fields, including insurance, finance, real estate, health services, government and human resources.

This post was originally published in the IAIDQ IDQ Newsletter, April 2005 Vol. 1, Issue 2 © 2005 ISO Properties, Inc.

About IAIDQ

Celebrating its 5th Anniversary this year, the IAIDQ is the only Professional Association focusing specifically on the needs and issues of information quality and data quality professionals. Lead and run by a team of volunteers (all of them practitioners or researchers in the Information Quality field), the IAIDQ runs on a not-for-profit basis to develop the profession and the skills and understanding of professionals. The Association was founded by Larry English and Tom Redman.

Among our key initiatives are the development of a vendor neutral certification for Information Quality professionals (the CIQP certification), the development of Communities of Practice in various areas of interest (geographic, industry based, knowledge-domain based), and generally working to raise awareness of the importance of good quality information as a critical business asset and the importance of managing it as such.

The Association has recently published the first ever Salary and Job Satisfaction study in the Information/Data Quality Profession, which examined the remuneration and satisfaction of IDQ professionals from around the world. That report can be downloaded from the IAIDQ Website (http://iaidq.org/publications/pierce-2009-07.shtml)

You can find us at: http://iaidq.org, or on Twitter at http://twitter.com/iaidq or on LinkedIn (where we are rolling out extensive supports for our Communities of Practice).